dividing groundtruth into training, testing and validation

Question

0 个投票

Hi, I have 800 images and labels in data set and I wanted to divide it into test, train and validation set.

I am using following code:

load gTruth % load gTruth file of training data

[imds,pxds] = pixelLabelTrainingData(gTruth);

imdsTrain=imds(81:720);

pxdsTrain=pxds(81:720);

imdsTest=imds(1:80);

pxdsTest=pxds(1:80);

imdsVal=imds(721:800);

pxdsVal=pxds(721:800);

its not working, any possible solution....?

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

Answer 1

yanqi liu 2021-2-5

0 个投票

[imds60, imds10, imds30] = splitEachLabel(imds,0.6,0.1)

Raza Ali 2021-2-5

While using this command, I am getting following error:

Error using matlab.io.datastore.ImageDatastore/splitEachLabel (line 211)

splitEachLabel requires non-empty Labels property.

yanqi liu 2021-2-5

can you upload the demo data, and we can make some test case.